Merge branch 'master' of https://github.com/higginsdragon/docx into higgins-merge

Conflicts:
	lib/docx/document.rb
	lib/docx/parser.rb
	spec/docx/document_spec.rb

Author: bitops

lib/docx/containers/paragraph.rb

@@ -15,9 +15,11 @@ def self.tag
 
         # Child elements: pPr, r, fldSimple, hlink, subDoc
         # http://msdn.microsoft.com/en-us/library/office/ee364458(v=office.11).aspx
-        def initialize(node)
+        def initialize(node, document_properties = {})
           @node = node
           @properties_tag = 'pPr'
+          @document_properties = document_properties
+          @font_size = @document_properties[:font_size]
         end
 
         # Set text of paragraph
@@ -39,17 +41,55 @@ def to_s
           text_runs.map(&:text).join('')
         end
 
+        # Return paragraph as a <p></p> HTML fragment with formatting based on properties.
+        def to_html
+          html = ''
+          text_runs.each do |text_run|
+            html << text_run.to_html
+          end
+          styles = { 'font-size' => "#{font_size}pt" }
+          styles['text-align'] = alignment if alignment
+          html_tag(:p, content: html, styles: styles)
+        end
+
+
         # Array of text runs contained within paragraph
         def text_runs
-          @node.xpath('w:r|w:hyperlink/w:r').map {|r_node| Containers::TextRun.new(r_node) }
+          @node.xpath('w:r|w:hyperlink/w:r').map { |r_node| Containers::TextRun.new(r_node, @document_properties) }
         end
 
         # Iterate over each text run within a paragraph
         def each_text_run
           text_runs.each { |tr| yield(tr) }
         end
+
+        def aligned_left?
+          ['left', nil].include?(alignment)
+        end
+
+        def aligned_right?
+          alignment == 'right'
+        end
+
+        def aligned_center?
+          alignment == 'center'
+        end
+
+        def font_size
+          size_tag = @node.xpath('w:pPr//w:sz').first
+          size_tag ? size_tag.attributes['val'].value.to_i / 2 : @font_size
+        end
 
         alias_method :text, :to_s
+
+        private
+
+        # Returns the alignment if any, or nil if left
+        def alignment
+          alignment_tag = @node.xpath('.//w:jc').first
+          alignment_tag ? alignment_tag.attributes['val'].value : nil
+        end
+
       end
     end
   end

lib/docx/containers/text_run.rb

@@ -20,12 +20,14 @@ def self.tag
         attr_reader :text
         attr_reader :formatting
 
-        def initialize(node)
+        def initialize(node, document_properties = {})
           @node = node
           @text_nodes = @node.xpath('w:t').map {|t_node| Elements::Text.new(t_node) }
           @properties_tag = 'rPr'
           @text       = parse_text || ''
           @formatting = parse_formatting || DEFAULT_FORMATTING
+          @document_properties = document_properties
+          @font_size = @document_properties[:font_size]
         end
 
         # Set text of text run
@@ -54,7 +56,20 @@ def parse_formatting
         def to_s
           @text
         end
-        
+
+        # Return text as a HTML fragment with formatting based on properties.
+        def to_html
+          html = @text
+          html = html_tag(:em, content: html) if italicized?
+          html = html_tag(:strong, content: html) if bolded?
+          styles = {}
+          styles['text-decoration'] = 'underline' if underlined?
+          # No need to be granular with font size down to the span level if it doesn't vary.
+          styles['font-size'] = "#{font_size}pt" if font_size != @font_size 
+          html = html_tag(:span, content: html, styles: styles) unless styles.empty?
+          return html
+        end
+
         def italicized?
           @formatting[:italic]
         end
@@ -66,6 +81,11 @@ def bolded?
         def underlined?
           @formatting[:underline]
         end
+
+        def font_size
+          size_tag = @node.xpath('w:rPr//w:sz').first
+          size_tag ? size_tag.attributes['val'].value.to_i / 2 : @font_size
+        end
       end
     end
   end

lib/docx/document.rb

@@ -1,6 +1,8 @@
-require 'docx/parser'
-require 'zip'
-
+require 'docx/containers'
+require 'docx/elements'
+require 'nokogiri'
+require 'zip/zip'
+
 module Docx
   # The Document class wraps around a docx file and provides methods to
   # interface with it.
@@ -16,17 +18,28 @@ module Docx
   #     puts d.text
   #   end
   class Document
-    delegate :paragraphs, :bookmarks, :tables, :to => :@parser
-    delegate :doc, :xml, :zip, :to => :@parser
+    attr_reader :xml, :doc, :zip, :styles
+    
     def initialize(path, &block)
       @replace = {}
+      @zip = Zip::ZipFile.open(path)
+      @document_xml = @zip.read('word/document.xml')
+      @doc = Nokogiri::XML(@document_xml)
+      @styles_xml = @zip.read('word/styles.xml')
+      @styles = Nokogiri::XML(@styles_xml)
       if block_given?
-        @parser = Parser.new(File.expand_path(path), &block)
-      else
-        @parser = Parser.new(File.expand_path(path))
+        yield self
+        @zip.close
       end
     end
-    
+
+    # This stores the current global document properties, for now
+    def document_properties
+      {
+        font_size: font_size
+      }
+    end
+
     # With no associated block, Docx::Document.open is a synonym for Docx::Document.new. If the optional code block is given, it will be passed the opened +docx+ file as an argument and the Docx::Document oject will automatically be closed when the block terminates. The values of the block will be returned from Docx::Document.open.
     # call-seq:
     #   open(filepath) => file
@@ -35,6 +48,30 @@ def self.open(path, &block)
       self.new(path, &block)
     end
 
+    def paragraphs
+      @doc.xpath('//w:document//w:body//w:p').map { |p_node| parse_paragraph_from p_node }
+    end
+
+    def bookmarks
+      bkmrks_hsh = Hash.new
+      bkmrks_ary = @doc.xpath('//w:bookmarkStart').map { |b_node| parse_bookmark_from b_node }
+      # auto-generated by office 2010
+      bkmrks_ary.reject! {|b| b.name == "_GoBack" }
+      bkmrks_ary.each {|b| bkmrks_hsh[b.name] = b }
+      bkmrks_hsh
+    end
+
+    def tables
+      @doc.xpath('//w:document//w:body//w:tbl').map { |t_node| parse_table_from t_node }
+    end
+
+    # Some documents have this set, others don't.
+    # Values are returned as half-points, so to get points, that's why it's divided by 2.
+    def font_size
+      size_tag = @styles.xpath('//w:docDefaults//w:rPrDefault//w:rPr//w:sz').first
+      size_tag ? size_tag.attributes['val'].value.to_i / 2 : nil
+    end
+
     ##
     # *Deprecated*
     # 
@@ -51,6 +88,11 @@ def to_s
       paragraphs.map(&:to_s).join("\n")
     end
 
+    # Output entire document as a String HTML fragment
+    def to_html
+      paragraphs.map(&:to_html).join('\n')
+    end
+
     # Save document to provided path
     # call-seq:
     #   save(filepath) => void
@@ -83,5 +125,18 @@ def update
       @replace["word/document.xml"] = doc.serialize :save_with => 0
     end
 
+    # generate Elements::Containers::Paragraph from paragraph XML node
+    def parse_paragraph_from(p_node)
+      Elements::Containers::Paragraph.new(p_node, document_properties)
+    end
+
+    # generate Elements::Bookmark from bookmark XML node
+    def parse_bookmark_from(b_node)
+      Elements::Bookmark.new(b_node)
+    end
+
+    def parse_table_from(t_node)
+      Elements::Containers::Table.new(t_node)
+    end
   end
 end

lib/docx/elements/element.rb

@@ -55,6 +55,30 @@ def copy
         self.class.new(@node.dup)
       end
 
+      # A method to wrap content in an HTML tag.
+      # Currently used in paragraph and text_run for the to_html methods
+      #
+      # content:: The base text content for the tag.
+      # styles:: Hash of the inline CSS styles to be applied. e.g.
+      #          { 'font-size' => '12pt', 'text-decoration' => 'underline' }
+      #
+      def html_tag(name, options = {})
+        content = options[:content]
+        styles = options[:styles]
+
+        html = "<#{name.to_s}"
+        unless styles.nil? || styles.empty?
+          styles_array = []
+          styles.each do |property, value|
+            styles_array << "#{property.to_s}:#{value};"
+          end
+          html << " style=\"#{styles_array.join('')}\""
+        end
+        html << ">"
+        html << content if content
+        html << "</#{name.to_s}>"
+      end
+
       module ClassMethods
         def create_with(element)
           # Need to somehow get the xml document accessible here by default, but this is alright in the interim